Turn file inode management completely over to the block processor

If the block processor allocates and dynamically resizes inodes on
the fly, we can add data indefinitely without knowing the size of
the file ahead of time.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

1 files changed, 13 insertions, 10 deletions

diff --git a/include/sqfs/block_processor.h b/include/sqfs/block_processor.h
index 6945633..2ad12c0 100644
--- a/include/sqfs/block_processor.h
+++ b/include/sqfs/block_processor.h
@@ -142,24 +142,27 @@ sqfs_block_processor_t *sqfs_block_processor_create(size_t max_block_size,
  * are done. After writing all files, use @ref sqfs_block_processor_finish to
  * wait until all blocks that are still in flight are done and written to disk.
  *
- * The specified inode pointer is kept internally and updated with the
- * compressed block sizes and final destinations of the file and possible
- * fragment. You need to make sure it has enough backing-store for all blocks
- * to come. Furthermore, since there can still be blocks in-flight even after
- * calling @ref sqfs_block_processor_end_file, the data in the inode may still
- * change. The only point at which the data writer is guarnteed to not touch
- * them anymore is after @ref sqfs_block_processor_finish has returned.
+ * The specified pointer to an inode pointer is kept internally and updated with
+ * the compressed block sizes and final destinations of the file and possible
+ * fragment. Since the inode may have to be grown to larger size, the actual
+ * inode pointer can change over time. Furthermore, since there can still be
+ * blocks in-flight even after calling @ref sqfs_block_processor_end_file, the
+ * data in the inode and the value of the pointer may still change. The only
+ * point at which the data writer is guaranteed to not touch them anymore is
+ * after @ref sqfs_block_processor_sync or @ref sqfs_block_processor_finish has
+ * returned.
  *
  * @param proc A pointer to a data writer object.
- * @param inode The regular file inode representing the file. The data writer
- *              internally updates it while writing blocks to disk.
+ * @param inode A pointer to a pointer to an inode. The block processor creates
+ *              a file inode and stores a pointer to it here and keeps updating
+ *              the inode as the file grows.
  * @param flags A combination of @ref E_SQFS_BLK_FLAGS that can be used to
  *              micro manage how the data is processed.
  *
  * @return Zero on success, an @ref E_SQFS_ERROR value on failure.
  */
 SQFS_API int sqfs_block_processor_begin_file(sqfs_block_processor_t *proc,
-					     sqfs_inode_generic_t *inode,
+					     sqfs_inode_generic_t **inode,
 					     sqfs_u32 flags);
 
 /**